[The picture is from InsuranceBUSINESS,the copyright belongs to the original author]

In the insurance industry, underwriting has always been a key business task. But the old way has many issues, like slow speed, poor risk spotting, high fraud chance, and a bad experience for customers. These problems not only hold back business growth but also hurt customer happiness. However, with the fast growth of artificial intelligence technology, there are new answers to these problems now.^[1]

Intelligent underwriting simply uses artificial intelligence and big-data tools to replace the old human review method. Before, underwriting usually meant people checked the applicant's details, like health, job, and age, to choose whether to give insurance and for how much. Now, intelligent underwriting handles this data more smoothly through automated risk checks and can quickly give a yes or no answer.

Challenges of traditional underwriting: Double pressure of efficiency and risk

Underwriting is the main way insurance companies earn money, but this task has always faced many problems. The old underwriting way depends on human skill and set rules, which can't deal with the fast-moving risk world and the need for exact pricing. ^[2]For instance, over 70% of the key data in insurance details is spread out in unorganized texts like doctor reports and money records. Old technology can only pull out less than 60% of the data, causing pricing mistakes of up to 15% - 20%. Also, fraud events in the business are rising by 23% each year, and the error rate of current anti-fraud tools is as high as 18%, leading to big losses for insurance companies every year. What's more, old human underwriting takes a long time, which badly hurts the customer experience.

Operation process of intelligent underwriting:

The process of intelligent underwriting is very straightforward. First, the applicant provides personal details, like age, gender, job, and health, using a website or an app. Then, the system uses AI and big data to analyze this information. This step involves organizing the data, looking for patterns, and checking for risk. Next, the system evaluates the applicant's risk by using a fixed model, often a machine learning algorithm. Finally, the system gives its insurance recommendation based on the risk results. This includes whether to insure, the amount of coverage, and the price. This whole process is fast and accurate.

[The picture is from Rapid Acceleration Partners,the copyright belongs to the original author]

Intelligent underwriting is a turning point for insurance companies:

Intelligent underwriting is a major change for insurance companies. The main reason for this is that the old underwriting method cannot meet today's fast-changing market needs. The old models used historical data and fixed rules to assess risk. But now, with new technology, climate changes, and different customer demands, the old method is not effective. Intelligent underwriting can analyze data in real time and predict future trends using AI and machine learning. So, it helps companies build risk models that are more flexible and forward-looking. As a result, insurance companies can manage risk more precisely, set better prices, and offer personalized and quick services. This leads to better customer relationships and a stronger position in the market.

Specific advantages of intelligent underwriting:

The artificial intelligence model can lower the math cost while keeping its strong power through a mixed expert system. It can grasp complex texts, work with many data forms, and even quickly study up to 128K of application papers, correctly pull out risk items, and give underwriting ideas right away.

Improve underwriting efficiency

Intelligent underwriting has made underwriting faster. In the past, human underwriting took an average of 48 hours, but now it only takes 15 minutes. Through many-sided data study, the model can raise the correctness of finding high-risk cases to 91%, so cutting the underwriting risk. Take health insurance as an example. The model can automatically match odd signs in health histories and exam reports, cutting the time for human checks from 20 minutes to 45 seconds, greatly raising the underwriting speed and accuracy.

Accurately identify risks

The artificial intelligence model can handle unorganized data, like doctor reports and scan results, and automatically pull out hidden risk items, like long-term illnesses and family diseases, with a correctness rate of up to 92%. It can also bring together data from many places, like smart devices and online networks, to build a more exact moving risk picture. For example, when setting prices for car insurance, the model can cut the loss guess error by 19.7% by studying many sides, like how a person drives, online data, and weather details.

Fraud detection

Artificial intelligence also does very well in finding fraud. Through knowledge web technology, the model can spot odd application acts, raising the find rate of fraud cases by 40% and keeping the false alarm rate under 5%. It can automatically find word patterns in insurance applications, match mismatches between health forms and health card records, find unclear data in business insurance applications, and spot repeat application acts, cutting fraud risks.

Reduce operating costs

After adding the large artificial intelligence model, insurance companies have greatly cut their running costs. The automated underwriting process has cut the sorting of copy papers by over 60% and saved about 40% of worker costs. Also, the exact risk check and fraud find skills have also effectively cut the claim rate and raised the underwriting profit. Guesses show that after full use, the underwriting profit margin can be raised by 2.3 percentage points, bringing major money gains.

Success stories

A large property insurance company has raised its car insurance underwriting speed by 40% by using an artificial intelligence model. Before, underwriting took 48 hours, but now it only takes 15 minutes, and the underwriting correctness has reached 92%. The fraud spot rate has gone up from 67% to 89%, and the human check rate has gone down from 100% to 8%, greatly lowering running costs and raising speed.

A life insurance company uses an artificial intelligence model to actively create health questions. It changes the depth of questions as they come based on the applicant's first answers, raising the form finish rate by 40% and cutting the bad choice risk by 18%. This method not only makes the customer experience better but also lowers risks.

A reinsurance company has used a large artificial intelligence model in disaster risk modeling. By joining weather data, place details, and past loss data, it has made a more correct risk check system. The loss guess error rate of this model is only one-third of the old way, helping the company save tens of millions in reinsurance costs.

[The picture is from Business Administration Information,the copyright belongs to the original author]

Challenges and solutions of intelligent underwriting for insurance companies:

Poor data quality

Many insurance companies use old data and paper records. This causes bad data quality. AI needs clean and standard data to work well.^[3] If data is in different places and formats, the AI will not work properly. For example, claim forms might be handwritten. Also, customer information can be in different departments. This bad data changes the AI's results. It also makes moving to digital systems slower.

Insurance companies should create a strong data management system. This makes sure data is cleaned and organized at every step. They can use main data management tools. They can also use explainable AI (XAI). This makes the decision process easier to see.

Outdated technology

Many insurance companies use old computer systems. These systems were not built for modern AI. Old systems do not have enough processing power. They cannot handle data in real time. This causes problems. For example, it is hard to add AI models. It can also cause delays or system breaks. Also, old systems cannot use data from new sources, like sensors.

Insurance companies can start with simple AI tasks. For example, they can sort documents or answer emails. This avoids major changes to old systems at first. After small projects work, they can then do more complex tasks. This includes finding fraud or automating underwriting.

Inaccurate content

Generative AI (GenAI) can create wrong information. It can also invent facts or break rules. This can happen when it writes insurance policies, runs chatbots, or makes ads. In insurance, using the right words is very important. Wrong AI content can confuse customers. It can also lead to breaking the law.

Insurance companies should use generative AI as a helper. It should not create content alone. Everything the AI makes must be reviewed by a person. This is especially important for customer materials. This check makes sure the content is correct and follows rules. Companies should limit what generative AI does. For example, use it to write summaries, not to make final decisions. A person must check every step.

Compliance and ethical risks

Using AI must be fair and clear. Its decisions must also be trackable. If an AI model's decision cannot be explained, the company could face lawsuits. It could also harm its reputation. For example, the model could be biased. This could cause unfair claim denials or pricing. Regulators are now stricter about automated decisions.

Insurance companies need to use explainable AI (XAI) tools. This ensures every decision has a clear reason. They must also check the model often for bias. They should work with their legal and compliance teams. This ensures all decision steps can be reviewed.

Source:

[1] ScienceDirect: “AI revolutionizing industries worldwide: A comprehensive overview of its diverse applications”

[2] Mckinsey: “Underwriting talent: Strategies for property and casualty insurers”

[3] simbo.ai “The Critical Importance of Data Governance in Ensuring Secure, Compliant, and Ethical AI Deployment in Healthcare Settings”

Reference：

[1]Invensis: “AI in Insurance: Key Benefits, Use Cases, and Challenges”

[2]Txminds: “Intelligent Underwriting: Moving Beyond Risk Assessment to Risk Advantage”